A Personalisable Electronic Book for Video-based Sign Language ...
A Personalisable Electronic Book for Video-based Sign Language ...
A Personalisable Electronic Book for Video-based Sign Language ...
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Ohene-Djan, J., Zimmer, R., Gorle, M., & Naqvi, S. (2003). A <strong>Personalisable</strong> <strong>Electronic</strong> <strong>Book</strong> <strong>for</strong> <strong>Video</strong>-<strong>based</strong> <strong>Sign</strong><br />
<strong>Language</strong> Education. Educational Technology & Society, 6 (4), 86-99, Available at http://ifets.ieee.org/periodical/6_4/9.pdf<br />
A <strong>Personalisable</strong> <strong>Electronic</strong> <strong>Book</strong> <strong>for</strong> <strong>Video</strong>-<strong>based</strong> <strong>Sign</strong> <strong>Language</strong> Education<br />
James Ohene-Djan, Robert Zimmer, Matthew Gorle and Saduf Naqvi<br />
Department of Computing<br />
Goldsmiths College, University of London<br />
New Cross, London, SE14 6NW, United Kingdom<br />
j.djan@gold.ac.uk<br />
r.zimmer@gold.ac.uk<br />
m.gorle@gold.ac.uk<br />
s.naqvi@gold.ac.uk<br />
Abstract<br />
The World Wide Web (WWW) is an increasingly important source of educational material. However, only<br />
scant research has been directed towards making this educational in<strong>for</strong>mation accessible to the deaf.<br />
Adaptive Hypermedia (AH) is an area of research that aims to enhance the functionality of hypermedia<br />
systems, such as the WWW, by enabling users to personalise their interaction with the digital content<br />
provided by such systems. In this paper, we present a personalisable electronic book, Kids <strong>Sign</strong> Online<br />
(KSO). KSO utilises AH techniques together with digital video content to assist in the teaching of British<br />
<strong>Sign</strong> <strong>Language</strong> to deaf children. By providing an example of an adaptive WWW-<strong>based</strong> sign language<br />
education system, we hope to broaden the accessibility of sign language learning materials and to illustrate<br />
how AH tools and techniques may be used to address issues related to the education of the deaf in online<br />
learning environments.<br />
Keywords<br />
Disability learning, Hypermedia, Adaptive hypermedia, Digital content, Online learning, Accessibility, <strong>Sign</strong><br />
language<br />
Introduction<br />
A sign language is a language that uses combinations of hand shapes and hand, arm and/or body movements<br />
together with facial expressions to communicate without using sound (Sutton-Spence R & Woll B, 1999). In the<br />
last thirty years, sign languages have become the standard way of teaching language to deaf children. These<br />
languages have enabled deaf children to communicate well with other people, both hearing and deaf. It is now<br />
clear that, using sign language, deaf children can have perfectly normal language and communication skills<br />
(Fischer R & Lane H, 1993). For people brought up with them, sign languages play the role of native natural<br />
languages, composed of intricate syntaxes and vocabularies.<br />
In the context of education, both spoken and printed in<strong>for</strong>mation has historically been translated into sign<br />
language, in order to enable deaf students to have equal access to it. Today, the World Wide Web (WWW), a<br />
highly interconnected collection of computational resources, is a major source of educational in<strong>for</strong>mation.<br />
School children routinely use the WWW <strong>for</strong> both in<strong>for</strong>mation seeking tasks and discovery learning. There has<br />
already been some work done towards tailoring web resources to increase their accessibility <strong>for</strong> the deaf. The<br />
main strands of the research are: how to represent sign languages within the digital domain, <strong>for</strong> example Visicast<br />
(Visicast, 2003) and Vsign (Vsign, 2003), and how to devise writing systems <strong>for</strong> sign languages utilising such<br />
media as line drawings, symbols and photographs (Kramer & Ovadia 2000). This paper concerns an attempt to<br />
use more sophisticated techniques <strong>for</strong> tailoring online material <strong>for</strong> the needs of speakers of sign languages. Our<br />
approach is to use Adaptive Hypermedia.<br />
Adaptive Hypermedia (AH) is a set of ideas and techniques aimed at extending hypermedia systems, such as the<br />
WWW, to enable a user’s interaction to be personalised on an individual basis (Brusilovsky P et al., 1998). AH<br />
systems use knowledge provided by, or inferred from the behaviour of, users to tailor what in<strong>for</strong>mation is<br />
presented to the user and in what <strong>for</strong>m. AH systems support users in navigating hyper-documents by limiting<br />
options <strong>for</strong> traversal, suggesting links to follow, and providing additional in<strong>for</strong>mation about links and other<br />
resources. AH systems tend to be particularly useful in areas such as learning, where users have differing<br />
in<strong>for</strong>mation seeking requirements and different histories and preferences.<br />
ISSN 1436-4522 (online) and 1176-3647 (print). © International Forum of Educational Technology & Society (IFETS). The authors and the <strong>for</strong>um jointly retain the<br />
copyright of the articles. Permission to make digital or hard copies of part or all of this work <strong>for</strong> personal or classroom use is granted without fee provided that copies<br />
are not made or distributed <strong>for</strong> profit or commercial advantage and that copies bear the full citation on the first page. Copyrights <strong>for</strong> components of this work owned by<br />
others than IFETS must be honoured. Abstracting with credit is permitted. To copy otherwise, to republish, to post on servers, or to redistribute to lists, requires prior<br />
specific permission and/or a fee. Request permissions from the editors at kinshuk@massey.ac.nz.<br />
86
In this paper, we present a personalisable electronic book, Kids <strong>Sign</strong> Online (KSO) (Naqvi, S et al, 2003)., that<br />
supports the learning of British <strong>Sign</strong> <strong>Language</strong> (Flodin M, 1994) to deaf children and their tutors. The system’s<br />
uniqueness lies in its use of personalisation and adaptation (P&A) techniques, incorporated with digital video<br />
content, in the development of WWW-<strong>based</strong> electronic books <strong>for</strong> children. The British Deaf Association states<br />
that:<br />
“Deaf people have the right to a quality education throughout their lives, which accepts their<br />
linguistic, cultural and social identity, which builds positive self-esteem and sets no limit to their<br />
learning.” (British Deaf Association, 2001).<br />
If such an aspiration is to be realised, we believe that it is crucial that educational materials found on the WWW<br />
are adapted to suit the needs of the deaf. The research reported in this paper is motivated by the view that<br />
through the application of P&A techniques, it will be possible to provide deaf students with equal access to<br />
WWW-<strong>based</strong> educational resources.<br />
Through the use of P&A techniques and digital sign-language content, KSO aims to assist deaf students in the<br />
individualised learning of sign language. Within the system, the teacher is realised as a composer and tailor of<br />
digital content that makes personalised digital video, textual descriptions and exercises available to students. By<br />
providing an adaptive WWW-<strong>based</strong> sign language education system, we hope to broaden the accessibility of<br />
sign language learning materials. Furthermore, we aim to illustrate how AH tools and techniques may be used to<br />
address issues related to the education of the deaf in online learning environments.<br />
The remainder of this paper is structured as follows. In section 2 we provide the motivations <strong>for</strong> our work in the<br />
areas of educational AH systems and sign language systems. Section 3 provides an overview of the Goldsmiths<br />
Adaptive Hypermedia Model, which provides the framework and architecture <strong>for</strong> KSO. In section 4, we<br />
describe how the system dynamically generates personalisable educational sign language material. Section 5<br />
describes how P&A features of KSO are implemented, and how they are made available to students. This<br />
section also describes features that enable students to search <strong>for</strong> personalised video content and describes the<br />
assessment subsystem used to provide students with exercises to assess their learning. Section 6 provides a brief<br />
overview of additional features that are found within KSO. These are an online sign language dictionary and<br />
digital video fairytale stories. Section 7 highlights related work and compares it to the work contributed by this<br />
paper. Finally, section 8 contains conclusions and future directions <strong>for</strong> research.<br />
Background and Motivations<br />
The potential <strong>for</strong> using AH systems to address issues of learning has long been recognised (Beaumont I &<br />
Brusilovsky P, 1995; Brusilovsky P et al, 1996). The predominant application area has been education (Kay J &<br />
Kummerfeld R J, 1994; Brusilovsky P et al, 1996; De Bra P, 1997; De Bra P, 1998). Early lab-<strong>based</strong> systems,<br />
such as these, were concerned with how AH techniques could be applied in an educational context. With the<br />
emergence of large-scale distributed hypermedia systems, such as the WWW, many online AH educational<br />
systems have been proposed (Weber G & Specht M, 1997; Steinacker A et al, 1998; Brusilovsky P, 1999; Stern<br />
M K & Woolf B P, 2000; Ng M H et al, 2002). Broadly, all these systems have a similar goal: to provide<br />
features that allow users to personalise, or have the system adapt, digital educational content. P&A features<br />
implemented by systems include: link hiding and annotation (Culver & De Bra P, 1997; Kay J& Kummerfeld B,<br />
1995), the incorporation of visual clues to assist learners (Brusilovsky P & Pesin L, 1995), the tailoring of<br />
content and its presentation (Specht & Oppermann, 1998; Ng M H et al, 2002), selective content (Ohene-Djan J<br />
& Fernandes AAA, 2002), and tailoring options <strong>for</strong> traversal through the inclusion of additional supplementary<br />
links (De Bra P & Calvi I, 1998). A review of recent AH concepts and issues can be found in (Brusilovsky P,<br />
2001).<br />
In this paper, we define personalisation to be the user-initiated tailoring of hyperdocuments. Adaptivity is<br />
defined to be system-driven personalisation.<br />
In order to understand how AH research can benefit the design of sign language educational systems, it is first<br />
useful to consider some of the reading, writing and teaching issues associated with the education of the deaf.<br />
Historically, deaf children were only taught to lip-read and speak (Fischer R & Lane H, 1993). As a result of<br />
their impediment, the reading levels of the deaf are generally of a lower standard than those of their hearing<br />
counterparts (Schleper D R & Mahshie S N, June 1997). Furthermore, those who are taught sign language as<br />
their primary mode of communication are potentially further disadvantaged when reading printed books or<br />
WWW-<strong>based</strong> textual content.<br />
87
A sign language is not a word-by-word translation of a spoken language, but is a language in its own right. <strong>Sign</strong><br />
languages, such as American <strong>Sign</strong> <strong>Language</strong> (Sternberg M L, 1996) and British <strong>Sign</strong> <strong>Language</strong> (Sutton-Spence<br />
R & Woll B, 1999), differ significantly in their syntax and semantics from American or British English.<br />
There<strong>for</strong>e, the relationship between written English and sign language is very different from the relationship<br />
between written English and spoken English. Written English is a <strong>for</strong>eign language to native speakers of sign<br />
languages. In sign languages movements, body positions and expressions are integral parts of the syntax. These<br />
do not easily translate to written texts. Furthermore, printed text often uses phraseology and grammatical<br />
constructs that make word by word translation difficult, if not impossible (Coleman J R & Wolf E E, 1991). As<br />
a result, printed text must nearly always be translated, by the teacher, into sign language, on behalf of the<br />
student, in real time. This process is expensive in terms of time and teaching resources and requires a high level<br />
of expertise.<br />
One approach to making printed, or WWW-<strong>based</strong>, educational materials accessible to deaf children is to<br />
establish a pictorial representation of the vocabulary and syntax of a sign language. These representations are<br />
called writing systems (Sutton V, 1996). Specifically, sign language writing systems use a set of visually<br />
designed symbols that record how people sign, thereby representing the visual subtleties of sign language.<br />
Although several proposals <strong>for</strong> sign language writing systems have been made (Sutton V, 1997), and tentative<br />
steps have been taken towards providing software-<strong>based</strong> sign language translators such as the <strong>Sign</strong>Writing<br />
Markup <strong>Language</strong>, SWML (da Rocha Costa A C, 2003), to allow <strong>for</strong> the automatic generation of electronic texts<br />
written in sign language, there is still no universally accepted writing system <strong>for</strong> sign languages.<br />
In the absence of a universally accepted writing system, video has emerged as the primary technological<br />
plat<strong>for</strong>m used to present educational material to the deaf. <strong>Sign</strong>ing books (Pragma, 2002) are video presentations<br />
of textual material. They are analogous to talking books <strong>for</strong> the blind. In such books, a narrator is recorded on<br />
videotape while signing the content of the book. The signing may be complemented by subtitles and other visual<br />
clues (i.e., graphics, animations, etc.). As signing books are presented via video, they inherit the inherent<br />
weaknesses of the medium. <strong>Video</strong>tape is inefficient to navigate through: a user must linearly shuttle backwards<br />
and <strong>for</strong>wards through the tape, as no direct access mechanism is available. Furthermore, users are unable to<br />
search <strong>for</strong> a specific phrase or topic. There are no contextual links <strong>for</strong> browsing and no table of contents to<br />
locate particular subjects. Finally, videotape degrades with repeated use. These weaknesses become particularly<br />
evident when such books are used <strong>for</strong> educational purposes.<br />
The deficiencies in videotape technology are a motivating factor <strong>for</strong> our use of hypermedia technologies as a<br />
plat<strong>for</strong>m <strong>for</strong> the delivery of sign language learning materials. The ability of hypermedia systems to make nonlinear<br />
context <strong>based</strong> navigation available to users, through the inclusion of hyperlinks, and their ability to make<br />
use of computational resources to dynamically manage content, suggests that this medium may be a more<br />
appropriate delivery plat<strong>for</strong>m. Furthermore, content accessible via hypermedia systems is stored digitally and is<br />
there<strong>for</strong>e not subject to degradation through repeated use.<br />
In the case of AH, we propose that users could benefit from possibilities, not only related to the medium, but to<br />
the design of the software itself. Adaptive hypermedia technologies are starting to prove effective in per<strong>for</strong>ming<br />
in<strong>for</strong>mation retrieval tasks when users have different in<strong>for</strong>mation seeking goals and histories (Karagiannidis C et<br />
al, 2001; Bajraktarevic N et al, 2003). The ability of these techniques to tailor the content made available on an<br />
individual basis and the ability to adapt this content to the specific knowledge that a user has of a subject could<br />
yield similar advantages to the deaf learner. Furthermore, the ability of adaptive hypermedia systems to tailor<br />
link presentation, in order to direct and assist users in their online learning, appears to make adaptive hypermedia<br />
techniques and technologies good candidates <strong>for</strong> use in the development of sign language education systems.<br />
KSO <strong>Electronic</strong> <strong>Book</strong> Framework and Systems Architecture<br />
In this section, we outline our view of electronic books, and provide an overview of the Goldsmiths Adaptive<br />
Hypermedia Model. This <strong>for</strong>mal model was used as the systems architecture in the design of the KSO adaptive<br />
electronic book.<br />
In this paper, we define the content of an <strong>Electronic</strong> <strong>Book</strong> (E-book) to be a hyperlinked network of digital<br />
in<strong>for</strong>mation units. When these in<strong>for</strong>mation units are rendered, they provide the user with optional links to other<br />
in<strong>for</strong>mation units. A rendering unit in an E-book is viewed as a hyperpage, and a collection of units as a<br />
hypermedia presentation. Such presentations are accessible via a WWW-browser (e.g., Galeon, Firebird,<br />
88
Internet Explorer, etc.). More generally, an E-book is viewed as an electronic version of a paperback book,<br />
whose digital content may be dynamically generated. This content is made accessible via the Internet, and is<br />
read using an E-book reader, such as a WWW-browser. In the case of KSO, an E-book reader is limited to be<br />
any device that is capable of displaying a WWW-page.<br />
Although there have been many impressive proposals <strong>for</strong> the development of E-book technologies (Harrison B<br />
L, 2000; Press L, 2000; Ohene-Djan J & Fernandes A A A, 2003). and much research has been conducted into<br />
their deployment (Landoni et al, 2000a) and design (Landoni et al, 2000b; Landoni et al, 2000c; Landoni et al,<br />
2001) a significant limitation of many WWW-<strong>based</strong> E-books is their inability to address users in<strong>for</strong>mation needs<br />
on an individual basis. For example, when a user interacts with a paper-<strong>based</strong> book, they are given the<br />
opportunity not only to read its content in a linear manner, but also to annotate it, insert and delete content, and<br />
to make, in context, cross references to parts of the book and other books. Such functionality is rarely found in<br />
E-books using the WWW and is a further motivating factor <strong>for</strong> the research reported in this paper. The KSO<br />
homepage is shown in Figure 1.<br />
Figure 1. The KSO Homepage<br />
An Overview of the Goldsmiths Adaptive Hypermedia Model (GAHM)<br />
The Goldsmiths Adaptive Hypermedia Model (GAHM) (Ohene-Djan J, 2000), (Ohene-Djan J & Fernandes A A<br />
A, 2000; Ohene-Djan J 2002; Ohene-Djan J and Fernandes A A A, 2002a) is a <strong>for</strong>mal characterisation of<br />
personalisation and adaptation within hypermedia systems. Its originality lies in its <strong>for</strong>mal specification of<br />
personalisable, adaptive hypermedia, when loosely coupled to user interface and database servers. Induced from<br />
a <strong>for</strong>mal definition of hypermedia specifications, used <strong>for</strong> the presentation of digital content, is a <strong>for</strong>mal model of<br />
hypermedia personalisation and adaptation.<br />
The model defines a rich set of user initiated personalisation actions, which enable individual users to<br />
personalise how in<strong>for</strong>mation seeking tasks are per<strong>for</strong>med. This is then extended with additional functionality to<br />
allow <strong>for</strong> the specification of a user model and decision making algorithm. These additional components allow<br />
<strong>for</strong> adaptivity to be realised. In GAHM, personalisation is viewed as the process of handing over to the user the<br />
ability to take actions to tailor hyperpages, thereby overriding aspects of a hyperpage’s content and presentation.<br />
Adaptivity is viewed as the process of allowing the system to take the initiative in tailoring actions in the light of<br />
its inference of a user’s in<strong>for</strong>mation seeking goals and history. In GAHM, adaptivity is, there<strong>for</strong>e, as expressive<br />
as personalisation, but requires the addition of user modelling and decision making technologies.<br />
Architecture of KSO<br />
Figure 2 depicts the open architecture <strong>for</strong> hypermedia systems (this, and all subsequent figures use classical<br />
UML notation), upon which KSO has been designed and implemented. This architecture reflects the<br />
89
predominant approach to the design of WWW-<strong>based</strong> hypermedia systems. A core of AH functionality is viewed<br />
as a client technology of one or more user interface and database servers (UISs and DBSs, respectively). In the<br />
case of KSO, the UISs are WWW-browsers, such as Galeon, Firebird and Internet Explorer, and the DBS used is<br />
PostGreSQL (Momjian B, 2001). The personalisation functionality implemented in KSO is represented by the<br />
shaded package.<br />
Figure 2. A General, Open Architecture <strong>for</strong> WWW-Based Hypermedia<br />
Broadly, once a request has been captured, from a user, <strong>for</strong> a hyperpage, it is channelled directly into the core of<br />
AH functionality. If the request is simply <strong>for</strong> a hyperpage to be shown, the core responds by composing a<br />
hyperpage specification and then rendering its digital content into a hypermedia presentation that can be<br />
displayed by a UIS. This composition process includes the querying of the PostGreSQL DBS, in order to fetch<br />
references to digital content. In KSO, hypermedia presentations consist of signing video clips, textual<br />
descriptions and assessment exercises. Other requests allow users to tailor hyperpage specifications and thereby<br />
personalise their hypermedia presentations. Such personalisation requests utilise meta-data, in the <strong>for</strong>m of<br />
annotations, which provide semantics <strong>for</strong> the content of hypermedia presentations.<br />
Hypermedia Presentations within KSO<br />
A hypermedia presentation is defined to be a collection of hyperpages, whose topology enables navigation<br />
between them. Within KSO, hyperpages are dynamically generated from hyperpage specifications. A<br />
hyperpage specification specifies a sequence of in<strong>for</strong>mation units, known as chunks. Chunks contain digital<br />
content, such as video clips, text and graphics. Each chunk is comprised of a specification of its content and a<br />
specification of how to present (or render) this content. Content specifications may be comprised of the content<br />
itself (as is the case with many of the textual descriptions in KSO) or a reference, in the <strong>for</strong>m of an SQL query,<br />
that, when executed, evaluates into content, as is the case with the signing digital video clips available in KSO.<br />
Figure 3. The Structure of a Hyperpage<br />
Rendering specifications define how the digital content, defined by a content specification, is to be presented by<br />
a UIS. They take the <strong>for</strong>m of a <strong>for</strong>mal text in a language that the UIS can render (e.g., HTML). There is a<br />
binding between the two specifications, which is achieved by interdispersing the renderable text with variables<br />
that reference the digital content of the content specification.<br />
90
Each chunk may be associated with a set of entry points and a set of exit points. An entry point enables the<br />
hyperpage where the chunk occurs to be referenced in a request. An exit point enables the chunk to establish a<br />
navigable link to a hyperpage denoted by the exit point. In WWW parlance, an entry point can be thought of as<br />
a Uni<strong>for</strong>m Resource Locator (URL), or as an anchor within a hyperpage, and an exit point as a hyperlink, e.g., an<br />
tag in HTML. The structure of a hyperpage is shown in Figure 3. A possible rendering of the chunk<br />
specification, <strong>for</strong> which a <strong>for</strong>mal language has been defined (Ohene-Djan J, 2000), is shown in Figure 4.<br />
Figure 4. A possible rendering of a Hyperpage Specfication<br />
Dynamically Generating Hypermedia Presentations<br />
When a user issues a request to view a hypermedia presentation via a UIS, the KSO system generates it by<br />
retrieving, in sequence, each of the hyperpage specifications <strong>for</strong> that presentation from a data store, referred to as<br />
a hyper-library (see Figure 5). A composition function then parses each hyperpage specification into a set of<br />
instructions that, when interpreted, assemble that part of the presentation as a renderable text. This interpretation<br />
process involves fetching digital content specified as an SQL query to the PostGreSQL DBS. Once this digital<br />
content is returned it is woven, using the binding mechanism of variables, into a renderable text. Once each<br />
hyperpage specification in the sequence has been composed the completed hypermedia presentation is returned<br />
to the UIS, which issued the original request. Figure 5 illustrates the dynamic generation of hypermedia<br />
presentations. A <strong>for</strong>mal model and associated semantics <strong>for</strong> the specification, and dynamic generation, of<br />
hyperpages can be found in Ohene-Djan J, 2000). Figure 6 shows an example of a hypermedia presentation<br />
containing assessment exercises in KSO.<br />
Figure 5. Dynamic Generation of Hypermedia Presentations<br />
91
Personalising Hypermedia Presentations<br />
Figure 6. A Test in KSO<br />
In KSO, the requests a user can issue to personalise hypermedia presentations are <strong>based</strong> on the annotating and<br />
rewriting of hyperpage specifications. An annotation pairs a hyperpage specification with notes of interest.<br />
These notes assign user-specific values to user-generic attributes of interest (e.g., The level of difficulty is high,<br />
the presentation medium is video, etc.). The existence of annotations allows <strong>for</strong> the personalisation of a<br />
specified chunk, a hyperpage, or a complete hypermedia presentation. Annotations are meta-data that describe<br />
the content and/or behaviour of a hyperpage or its component parts.<br />
In KSO, both hyperpages and their associated annotations are stored in the hyperlibrary. The hyperlibrary is<br />
implemented as a database (using the PostGreSQL relational database management system), comprised of<br />
relations over the relational schemas <strong>for</strong> hyperpages and hyperpage annotations. Further details on the<br />
modelling of the hyperlibrary can be found in (Ohene-Djan J, 2000).<br />
Personalisation Engine<br />
We view personalisation as the process of handing over to the user the ability to take actions to tailor hyperpages<br />
and, there<strong>for</strong>e, hypermedia presentations. The personalisation engine is initially set using a static user-model<br />
(Fischer G, 2000), which is a representation of a user’s knowledge of sign language knowledge. This is initially<br />
set, as a result of a student’s answers to a series of sign language exercises, when a user first registers with the<br />
system. This user model provides the means <strong>for</strong> a user to feed preferences into the personalisation engine.<br />
These details are then used to tailor which hypermedia content is made available <strong>for</strong> further personalisation by<br />
the user.<br />
Personalisation in KSO is implemented as a software mechanism that is responsible <strong>for</strong> providing the<br />
functionality required to tailor hypermedia presentations. Such tailoring causes a personalised version of a<br />
hypermedia presentation to be created. The dynamics of the personalisation mechanism may be understood as<br />
follows. The personalisation process starts when a user issues a personalisation request, via a UIS, to KSO. A<br />
personalisation request is comprised of a scope, which denotes the collection of hyperpages in the hyperlibrary<br />
<strong>for</strong> which the request should apply, and an action list, containing the tailoring actions to be executed over the<br />
scope. The interface used to issue such requests is shown in Figures 7 and 8. The example personalisation<br />
request shown in Figure 7 may be represented in natural language as “I wish to be presented with all pages that<br />
contain the string ‘1997’; subsequently delete the second chunk of each of these pages.” Figure 8 can be<br />
represented as “I wish to be presented with all pages in the hyperlibrary. Of these pages, only keep those that<br />
contain the string ‘1998’ in the second chunk.”<br />
92
On receiving this request, the composer and tailoring engine parses it to determine which hyperpages should be<br />
personalised and how, thereby generating a personalisation program. In the case of KSO, this program is a<br />
sequence of SQL statements, which are subsequently executed over the content of the hyperlibrary, thereby<br />
realising a state transition, in the hyperlibrary, that reflects the intended meaning of the personalisation request.<br />
KSO signing dictionary and fairytales<br />
Figure 7. Personalisation Request #1<br />
Figure 8. Personalisation Request #2<br />
During preliminary research conducted in conjunction with the British Deaf Association (BDA), Frank Barnes<br />
(Barnes, 2003) school <strong>for</strong> the deaf and Hawkswood (Hawkswood, 2003) School <strong>for</strong> the deaf it became apparent<br />
that there is a significant lack of BSL online dictionaries <strong>for</strong> deaf children. Furthermore, there are few<br />
educational materials, in the <strong>for</strong>m of digital video content, that address the learning needs of children under the<br />
age of ten years available via the WWW. As a result of these observations, it was decided that KSO would be<br />
designed to be bi-lingual (i.e., supporting the education of English as well as BSL) and would incorporate an<br />
online dictionary <strong>for</strong> the deaf and educational materials targeted towards children under the age of ten.<br />
KSO <strong>Sign</strong>ing Dictionary<br />
Although the BDA is currently working on the development of a large scale British dictionary <strong>for</strong> sign language,<br />
this dictionary is targeted at deaf adults. KSO incorporates an online dictionary that has been designed primarily<br />
<strong>for</strong> use by children. The KSO dictionary (Shown in Figure 9, 10, 11 and 12) enables a child to select a letter of<br />
the alphabet or chose a category (i.e., Animals, Food etc). As a result of this action the KSO displays a<br />
hypermedia presentation comprised of a selection of words which are within that category or which begin with<br />
the letter chosen, together with associated digital video, signing each word. Figures 9, 10, 11 and 12 show<br />
examples of the KSO online dictionary in operation<br />
93
Figure 9. KSO’s Online Dictionary<br />
Figure 10. KSO’s Online Dictionary<br />
Figure 11. KSO’s Online Dictionary<br />
94
<strong>Sign</strong>ing fairytales<br />
Figure 12. KSO’s <strong>Sign</strong>ing Dictionary<br />
A key feature in the design strategy <strong>for</strong> KSO was that the system’s content, as far as possible, reflected the<br />
learning needs of deaf children up to the age of 10. There<strong>for</strong>e, a decision was made that educational materials, in<br />
the <strong>for</strong>m of assessment exercises, would be supplemented by further materials that facilitated discovery-<strong>based</strong><br />
learning and general browsing <strong>for</strong> this age group. An example of this is the inclusion of a section comprised of<br />
classical children’s fairy tales, in the <strong>for</strong>m of signed digital video. It was felt that such content was important<br />
because it reflected the type of learning materials offered to hearing children of this age group. Each fairy tale is<br />
presented as both textual descriptions and signed digital video. An example of a fairy tale as a hypermedia<br />
presentation is given in Figure 13.<br />
Related Work<br />
Figure 13. Fairy Tales in KSO<br />
This section compares and contrasts the work reported here with that of other online sign language learning<br />
resources and that of AH systems in general. It is in no way exhaustive, but aims to give the reader a sense of<br />
where future technological developments may lie. As stated in the Background and Motivations section, the vast<br />
95
majority of deaf learning materials are paper, or videotape (Allsop L & Mason C), <strong>based</strong>. Some work has been<br />
done on WWW-<strong>based</strong> learning materials <strong>for</strong> the deaf (Deafchild International; Lapiak J A, 1996; Gaertner S,<br />
2003). Although these systems are valuable educational resources, they do not address users on an individual<br />
basis. Instead, they provide all users with the same in<strong>for</strong>mation, regardless of an individual user’s ability, or<br />
knowledge.<br />
The KSO system utilises digital video clips to represent sign language in the digital domain. Recent research<br />
ef<strong>for</strong>ts have aimed to incorporate 3D animations of signers in contrast to video. Vsign (Vsign, 2003), part of the<br />
MultiReader project (www.multireader.org), is a project consisting of a builder to create 3D animated avatars<br />
that per<strong>for</strong>m sign language motions, together with a player <strong>for</strong> such avatars. The Vsign builder allows users to<br />
build animations from a model of a human. Users may specify the complete range of body movements used in<br />
sign language. However, it is still not possible to include facial expressions. VisiCast (Visicast, 2003) is a<br />
project to develop a 3D model of a signing person to accompany subtitles and teletext in<strong>for</strong>mation on digital<br />
television. <strong>Sign</strong>s are digitally captured via advanced technologies, including a body suit and digital helmet. The<br />
VisiCast system is an extension of earlier work on the development of “Simon”, a prototype system that signed<br />
in English, as opposed to BSL. Although the VisiCast system uses 3D animations, these are displayed in 2D<br />
<strong>for</strong>m <strong>for</strong> per<strong>for</strong>mance reasons.<br />
Vcom3D (www.signingavatar.com) is currently developing leading edge software <strong>for</strong> the teaching of reading to<br />
the deaf. Their signing avatars, developed using 3D Studio (Murdock K, 2001), allow users to choose the<br />
perspective and signing speed. A particular point of note is Vcom3D’s ultimate goal; to develop a machine<br />
translation system that will take a website and output an animated, signed website.<br />
Although the authors of this paper accept that 2D representations may not completely represent all aspects of<br />
human signing that may be captured by a 3D model, it was felt that the benefits associated with digital video<br />
outweighed the costs associated with the construction of a 3D model. In particular, the use of digital video<br />
enabled KSO to show digital signing presentations by deaf children to deaf children. This, it was felt, is an<br />
important aspect of learning by example. The capture of digital video is, at present, relatively cheap in terms of<br />
time and cost, compared with that required to produce, and subsequently animate, 3D characters, such as those<br />
used by Vcom3D, VisiCast, and Vsign. Furthermore, the use of digital video, in contrast to 3D animations,<br />
enabled KSO to be available to a wide base of users, many of whom we found, during preliminary research, did<br />
not have the latest internet technologies available to them (i.e., up to date players, such as Flash and Shockwave,<br />
and the high per<strong>for</strong>mance hardware required to support these players).<br />
KSO’s use of digital video in the development of sign language dictionaries may be compared to that taken by<br />
the Personal Communicator (http://commtechlab.msu.edu/index.php), a CDROM-<strong>based</strong> ASL dictionary.<br />
Originally developed using Hypercard (Goodman D, 1998) at Michigan State University, this, now<br />
commercially available, software makes available to users over 2500 signs as digital video clips. In addition to<br />
ASL, the Personal Communicator also provides English synonyms. The Personal Communicator, now<br />
developed using Macromedia Director, is available <strong>for</strong> both the Macintosh and Windows plat<strong>for</strong>ms.<br />
The personalisation features of KSO are similar to those made available by implemented AH systems such as<br />
Adaptive Hyperman (Mathe N & Chen J, 1996), Joint Zone (Ng M H et al, 2002), and the AHA (De Bra P &<br />
Calvi L, 1998). These personalisation features include customisation via a static user model, the insertion and<br />
deletion of digital content, and the ability to express searches which result in tailored hypermedia presentations.<br />
Since the GAHM (Ohene-Djan J, 2000), the <strong>for</strong>mal model upon which KSO is <strong>based</strong>, characterises a complete<br />
set of possibilities <strong>for</strong> personalisation actions on WWW-<strong>based</strong> hyperdocuments, it is possible to implement all<br />
P&A actions described in (Brusilovsky P, 1996) and many of those described in (Brusilovsky P, 2001). It is<br />
envisaged that in future versions of KSO a broader range of P&A actions will be implemented.<br />
A significant difference between KSO and other AH systems is its application domain, namely sign language<br />
education <strong>for</strong> deaf children. Due to KSO’s application domain, its design incorporates several usability features<br />
not generally found in AH systems. For example, KSO allows children to customise their interface, in terms of<br />
colours and fonts. It also enables children to personalise pages with the child’s chosen identity.<br />
Finally, the design aspects of the hypermedia presentations contained within KSO, and their usability features,<br />
are derived in part, and shared by those of (Landoni et al, 2000b). However, in KSO, we use personalisation as a<br />
value added strategy, in contrast to, <strong>for</strong> example, visual clues (Landoni et al, 2000c).<br />
96
Conclusions<br />
To date, relatively little work has been directed towards providing online educational resources <strong>for</strong> the deaf,<br />
although the WWW is now a major source of educational in<strong>for</strong>mation <strong>for</strong> the hearing. We believe, that <strong>for</strong> deaf<br />
children to have quality education throughout their lives, in<strong>for</strong>mation contained within the WWW must be<br />
tailored to address their accessibility needs. The research reported in this paper is motivated by the view that,<br />
through the application of P&A techniques, it may be possible to provide deaf students with equal access to<br />
WWW-<strong>based</strong> educational resources.<br />
In this paper, we present KSO, a personalisable electronic book <strong>for</strong> the teaching of British <strong>Sign</strong> <strong>Language</strong> to deaf<br />
children and their tutors. The novelty of KSO is that it incorporates advanced P&A techniques into the<br />
development of a WWW-<strong>based</strong> electronic book. Through the use of the P&A techniques employed during the<br />
development of KSO, we aim to broaden the accessibility of online sign language learning materials. In this<br />
paper, we have illustrated how, through a combination of digital video, textual descriptions and assessment<br />
exercises, it is possible to devise an individualised learning environment <strong>for</strong> the deaf.<br />
KSO is a demonstration system, whose architecture is <strong>based</strong> upon a complete <strong>for</strong>malisation of P&A actions in<br />
hypermedia systems. We aim, through future development, to test the effectiveness of various P&A actions in<br />
the application area of deaf online learning.<br />
References<br />
Allsop, L. and Mason, C., (2002) A two-hour introduction to British <strong>Sign</strong> <strong>Language</strong> presented by Lorna Allsop<br />
and Clive Mason, British Deaf Association <strong>Video</strong> Service, Classification BSL9<br />
Bajraktarevic, N. Hall, W. & Fullick, P., (2003), Incorporating learning styles in hypermedia environment:<br />
Empirical evaluation, In proceedings of the Fourteenth Conference on Hypertext and Hypermedia Nottingham,<br />
UK, August 26, 2003.<br />
Barnes, (2003), Frank Barnes Primary School <strong>for</strong> Deaf Children, Harley Road, London.<br />
Beaumont, I. & Brusilovsky, P., (1995) Adaptive educational hypermedia: From ideas to real systems. In H.<br />
Maurer (Eds.), Proceedings of ED-MEDIA'95 - World conference on educational multimedia and hypermedia,<br />
Graz, Austria, June 17-21, 1995. Charlottesville, AACE, 93-98<br />
British Deaf Association, (2001), Education and Deaf People, leaflet.<br />
Brusilovsky, P. & Pesin, L. (1995). Visual Annotation of Links in Adaptive Hypermedia. In Proceedings of<br />
ACM CHI'95 Conference on Human Factors in Computing Systems, volume 2 of Short Papers: Agents and<br />
Anthropomorphism, pages 222-223, 1995.<br />
Brusilovsky, P., (1996) Methods and techniques of adaptive hypermedia. User Modeling and User-Adapted<br />
Interaction, 6 (2-3), 87-129.<br />
Brusilovsky, P. Schwarz, E. & Weber, G., (1996) A Tool <strong>for</strong> Developing Adaptive <strong>Electronic</strong> Textbooks on<br />
WWW. In Proceedings of WebNet'96, San Francisco, CA, AACE, 64-69.<br />
Brusilovsky, P. Schwarz, E. & Weber, G., (1996) ELM-ART: An intelligent tutoring system on World Wide<br />
Web. In C. Frasson, G. Gauthier, & A. Lesgold (Eds.), Third International Conference on Intelligent Tutoring<br />
Systems, ITS-96 (Lecture Notes in Computer Science, Vol. 1086), Berlin: Springer Verlag, pp. 261-269.<br />
Brusilovsky, P., Kobsa, A., & Vassileva J, (1998) editors Adaptive Hypertext and Hypermedia. Kluwer<br />
Academic Publishers, Dordrecht,. ISBN 0-7923-4843-5.<br />
Brusilovsky, P., (1999) Adaptive and Intelligent Technologies <strong>for</strong> Web-<strong>based</strong> Education In: C. Rollinger and C.<br />
Peylo (eds.) Künstliche Intelligenz, Special Issue on Intelligent Systems and Teleteaching, 1999, 4, 19-25.<br />
Brusilovsky, P. (2001) Adaptive hypermedia. User Modeling and User Adapted Interaction, Ten Year<br />
Anniversary Issue (Alfred Kobsa, ed.) 11 (1/2), 87-110<br />
Calvi, L., De Bra, P., (1997), Improving the Usability of Hypertext Courseware through Adaptive Linking,<br />
Proceedings of the Eighth ACM Conference on Hypertext, pp. 224-225, Washington DC, 1997.<br />
97
Coleman, J. R. & Wolf, E. E., (1991). Advanced <strong>Sign</strong> <strong>Language</strong> Vocabulary: A Resource Text <strong>for</strong> Educators,<br />
Interpreters, Parents, and <strong>Sign</strong> <strong>Language</strong> Instructors. ISBN: 0398057222<br />
De Bra, P., (1997). Teaching Through Adaptive Hypertext on the WWW. International Journal of Educational<br />
Telecommunications 3 (2), 163-179.<br />
De Bra, P., (1998), "2L690: Hypermedia Structures and Systems", Course at the Eindhoven University of<br />
Technology, 1994-1998. http://wwwis.win.tue.nl/2L690/<br />
De Bra, P. & Calvi, L., (1998), AHA! An open Adaptive Hypermedia Architecture. The New Review of<br />
Hypermedia and Multimedia 4, 115-139.<br />
Da Rocha Costa, A. C., (2003), SWML <strong>Sign</strong>Writing Markup <strong>Language</strong> http://swml.ucpel.tche.br/<br />
Deaf Child International Website, http://www.deafchild.org<br />
Fischer, G., (2000) User Modeling in Human-Computer Interaction, User Modeling in Human-Computer<br />
Interaction, User Modeling and User-Adapted Interaction, Kluwer Academic Publishers.<br />
Fischer, R. & Lane, H., (1993), Looking Back: A Reader on the History of Deaf Communities and Their <strong>Sign</strong><br />
<strong>Language</strong>s (International Studies on <strong>Sign</strong> <strong>Language</strong> and Communication of the). ISBN: 3927731323<br />
Flodin, M., (1994). <strong>Sign</strong>ing Illustrated: The Complete Learning Guide. ISBN: 0399521348<br />
Gaertner, S., (2003), ASL Fingerspelling, http://where.com/scott.net/asl/<br />
Goodman, D., (1998) The Complete Hypercard 2.2 Handbook Vol. 1, ToExcel; 4 th Edition, ISBN: 0966551427<br />
Harrison, B. L., (2000). E-books and the future of reading. IEEE Computer Graphics and Applications,<br />
20(3):32–39.<br />
Hawkswood, (2003), Hawkswood School, Antlers Hill, Ching<strong>for</strong>d, London<br />
Karagiannidis, C. Sampson, D. & Brusilovsky, P., (2001) "Empirical Evaluation of User Models and User-<br />
Adapted Systems: A Layered Evaluation Review", 8th Panhellenic Conference on In<strong>for</strong>matics, Nicosia, Cyprus,<br />
November 8-10, 2001, pp. 102-107.<br />
Kay, J. & Kummerfeld, B., (1994), An Individualised Course <strong>for</strong> the C Programming <strong>Language</strong>, in the<br />
proceedings of the second international WWW conference ’94, Mosaic and the Web Urbana IL 61801, USA<br />
Kramer & Ovadia (2000). You Can Learn <strong>Sign</strong> <strong>Language</strong>!: More Than 300 Words in Pictures (Winner of the<br />
National Parenting Association Gold Medal). ISBN: 0816763364<br />
Landoni, M. Crestani, F. & Melucci, M., (2000a). The visual book and the hypertextbook: Two electronic books<br />
one lesson? In Proceedings of the RIAO 2000 Conference, pages 247–265, Paris, France.<br />
Landoni, M. Wilson, R. & Gibb, F., (2000b). From the visual book to the web book: the importance of good<br />
design. In Proceedings of the Fourth European Conference on Research and Advanced Technology <strong>for</strong> Digital<br />
Libraries (ECDL 2000), pages 18–20.<br />
Landoni, M. Wilson, R. & Gibb, F., (2000c). From the visual book to the WEB book: The importance of good<br />
design. Lecture Notes in Computer Science, 1923:305–314.<br />
Landoni, M. Wilson, R. & Gibb, F., (2001). Looking <strong>for</strong> guidelines <strong>for</strong> the production of electronic textbooks.<br />
Online In<strong>for</strong>mation Review, 25(3).<br />
Lapiak, J. A., (1996), Handspeak: Visual Langauges, http://www.handspeak.com<br />
Mathe, N. & Chen, J., (1996). User-centered Indexing <strong>for</strong> Adaptive In<strong>for</strong>mation Access. User Modeling and<br />
User-Adapted Interaction, 6(2-3):225-261.<br />
Momjian, B., (2001) PostgreSQL introduction and concepts, ISBN 0-201-70331-9, Addison Wesley Professional<br />
Murdock, K., (2001), 3ds max 4 Bible, John Wiley & Sons, 2001, ISBN: 0764535846<br />
Naqvi, S., Ohene-Djan, J. & Gorle, M., (2003), Kids <strong>Sign</strong> Online, a public demo of which is available at<br />
http://doc.gold.ac.uk/~ma001sn<br />
98
Ng, M. H., Maier, P., Armstrong, R. & Hall, W., (2002) Making Web-<strong>based</strong> Learning Adaptive. In Barker,<br />
Philip and Rebelsky, Samuel, Eds. Proceedings ED-MEDIA 2002, World Conference on Educational<br />
Multimedia, Hypermedia & Telecommunication, pages 1436-1441, Denver, Colorado, USA.<br />
Ohene-Djan, J., (2000). A Formal Approach to <strong>Personalisable</strong>, Adaptive Hyperlink-Based Interaction. PhD<br />
thesis, Department of Mathematical and Computing Sciences, Goldsmiths College, University of London,<br />
University of London. Available from http://homepages.gold.ac.uk/johene-djan/pa-hlbs.ps.zip.<br />
Ohene-Djan, J. & Fernandes, A. A. A., (2000) A personalization framework <strong>for</strong> advanced learning technology.<br />
In Kinshuk, Jesshope C, and Okamoto T (Eds.), IEEE International Workshop on Advanced Learning<br />
Technologies (IWALT 2000), pages 21–24, Los Alamitos: IEEE Computer Society.<br />
Ohene-Djan, J., (2002) Ownership transfer via personalisation as a value-adding strategy <strong>for</strong> web-<strong>based</strong><br />
education. In P. Brusilovsky, N. Henze, and E. Millán (Eds.), Proceedings of the Workshop on Adaptive Systems<br />
<strong>for</strong> Web-<strong>based</strong> Education, 2nd International Conference on Adaptive Hypermedia and Adaptive Web Based<br />
Systems, Malaga, Spain, May 28 2002.<br />
Ohene-Djan, J. & Fernandes, A. A. A.,.(2002a) Modelling <strong>Personalisable</strong> Hypermedia: The Goldsmiths Model.<br />
New Review of Hypermedia and Multimedia, 8: 99–137.<br />
Ohene-Djan, J. & Fernandes, A. A. A., (2003). <strong>Personalisable</strong> electronic books. Journal of Digital In<strong>for</strong>mation,<br />
3(4), February 2003. Available from http: //jodi.ecs.soton.ac.uk/Articles/v03/i04/Ohene-Djan/.<br />
Pragma, N. L., (2002) <strong>Sign</strong>ing <strong>Book</strong>s <strong>for</strong> the Deaf http://www.signingbooks.org/<br />
Press, L., (2000). Personal computing: from P-books to E-books. Communications of the ACM, 43(5):17–17.<br />
Schleper, D. R. & Mahshie, S. N., (1997), Reading to Deaf Children: Learning from Deaf Adults (A Manual <strong>for</strong><br />
Parents and Teachers). ISBN: 0880952121<br />
Specht, M. & Oppermann, R., (1998), ACE - Adaptive Courseware Environment. The New Review of<br />
Hypermedia and Multimedia 4: 141-162 (1998)<br />
Steinacker, A., Seeberg, C., Rechenberger, K., Fischer, S. & Steinmetz, R., (1998). Dynamically generated tables<br />
of contents as guided tours in adaptive hypermedia systems. Educational Multimedia/Hypermedia and<br />
Telecommunications, Seattle, WA, AACE.<br />
Stern, M. K. & Woolf, B. P., (2000). Adaptive content in an online lecture system. In P. Brusilovsky, O. Stock,<br />
& C. Strapparava (Ed.), Adaptive Hypermedia and Adptive Web-<strong>based</strong> systens (pp. 225-238). Berlin: Springer-<br />
Verlag.<br />
Sternberg , M. L., (1996). Essential ASL : The Fun, Fast, and Simple Way to Learn American <strong>Sign</strong> <strong>Language</strong><br />
[ABRIDGED]. ISBN: 0062734288<br />
Sutton, V., (1996). <strong>Sign</strong> Writing Site. http://www.signwriting.org/<br />
Sutton, V., (1997). Dance Writing Site. http://www.dancewriting.org/<br />
Sutton-Spence, R. & Woll, B., (1999). The Linguistics of British <strong>Sign</strong> <strong>Language</strong> : An Introduction . ISBN:<br />
0521631424<br />
Visicast, (2003) Visicast http://www.visicast.co.uk<br />
Vsign, (2003) Vsign http://www.vsign.nl/<br />
Weber, G. & Specht, M., (1997) User Modeling and Adaptive Navigation Support in WWW-Based Tutoring<br />
Systems, In Anthony Jameson, Cécile Paris, and Carlo Tasso (Eds.) User Modeling: Proceedings of the Sixth<br />
International Conference, UM97 (pp. 289-300). Vienna, New York: Springer Wien New York.<br />
99